Reviewing those career trajectory predictions

Back in January, I played around with R and modeled some career trajectories for the White Sox players over 30.  Now that the season it over, it's time to review how those predictions turned out.

Paul Konerko was the first I looked at.  The model came up with a predicted OPS of .686.  After struggling earlier on, Konerko got his OPS all the way up to .683 on August 2.  That seems to be about when the wheels came off.  Konerko went 6 for 59 with four walks and a plunk the rest of the season to finish the season with an OPS of .572.

For Adam Dunn, the model predicted a very miserable .626 OPS.  There were problems with the model though.  Dunn's 2011 season was such an aberration that it seemed to break the predictive ability of the model.  Looking at his previous five season or so, Dunn did have a fairly linear drop of 30 to 40 points of OPS.  That would have put Dunn to around .722 to .732.  He actually looked like he was going to break the model completely when Dunn finish April with a .915 OPS.  As the season wore on, Dunn's OPS dropped to .773 when he was traded to Oakland.  Dunn finished the season at .752, only dropping 20 OPS points this season.

Alexei Ramirez actually had the dourest of OPS predictions for this season with a paltry .586 OPS.  Fortunately, he easily beat that with a .713 OPS.  Like Koneko, though, there is more to this story.  After starting out with a .910 OPS in April, his OPS was .669 for the rest of the season.  This is was right between the Steamer and Oliver projections prior to the season.

Finally, we come to Jeff Keppinger.  Keppinger's OPS seemed so random that there seemed to be no good way to predict his OPS.  Obviously, I couldn't predict that he'd start the season on the DL, get cut, and then have Rick Hahn thank Jerry Reinsdorf for letting him cut Keppinger.